# Financial Question Answering on FinVQA text-only (test) Source: [Financial Question Answering on FinVQA text-only (test)]() Collection index: [llms.txt]() Data updated: 2026-05-14T06:14:54.228Z Evaluation of model performance on the text-only test split of the FinVQA dataset for Financial Question Answering, reporting scores across multiple languages (English, Hindi, Bengali, Marathi, Gujarati, Tamil). A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Financial Question Answering | | Dataset (including any stated split) | FinVQA text-only (test) | | Metric | Accuracy (English) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 27 of 27 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Accuracy (English)](): Higher is better; unit: PERCENT. - [Accuracy (Hindi)](): Higher is better; unit: PERCENT. - [Accuracy (Bengali)](): Higher is better; unit: PERCENT. - [Accuracy (Marathi)](): Higher is better; unit: PERCENT. - [Accuracy (Gujarati)](): Higher is better; unit: PERCENT. - [Accuracy (Tamil)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (English) | Result date | Report paper | | --- | --- | --- | --- | | Qwen3-VL-32B-Instruct | 77.8 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen3-VL-32B-Instruct | 77.2 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen3-VL-32B-Instruct | 76.9 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen3-VL-8B-Instruct | 74.1 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen2.5-VL-32B-Instruct | 74.1 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen2.5-VL-32B-Instruct | 73.8 | Not recorded | 6a055748dbc836b87d828f1d | | Gemma-3-27b-it | 73.7 | Not recorded | 6a055748dbc836b87d828f1d | | Gemma-3-12b-it | 71.7 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen2.5-VL-32B-Instruct | 70.5 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen3-VL-8B-Instruct | 69.8 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen3-VL-4B-Instruct | 69.6 | Not recorded | 6a055748dbc836b87d828f1d | | Gemma-3-27b-it | 69.2 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen3-VL-8B-Instruct | 69 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen3-VL-4B-Instruct | 67.7 | Not recorded | 6a055748dbc836b87d828f1d | | Gemma-3-27b-it | 67.7 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen3-VL-4B-Instruct | 67.2 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen2.5-VL-3B-Instruct | 63.1 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen2.5-VL-3B-Instruct | 62.6 | Not recorded | 6a055748dbc836b87d828f1d | | Gemma-3-12b-it | 62.3 | Not recorded | 6a055748dbc836b87d828f1d | | Gemma-3-12b-it | 62.3 | Not recorded | 6a055748dbc836b87d828f1d | | Gemma-3-4b-it | 61.5 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen2.5-VL-7B-Instruct | 61.2 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen2.5-VL-7B-Instruct | 61.1 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen2.5-VL-3B-Instruct | 57.5 | Not recorded | 6a055748dbc836b87d828f1d | | Gemma-3-4b-it | 56.3 | Not recorded | 6a055748dbc836b87d828f1d | | Gemma-3-4b-it | 56.3 | Not recorded | 6a055748dbc836b87d828f1d | | Qwen2.5-VL-7B-Instruct | 55.6 | Not recorded | 6a055748dbc836b87d828f1d | ### Qwen3-VL-32B-Instruct Accuracy (English): 77.8. Result ID: 6a05681147f34ce924ebaac7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Zero shot inference"}. This record does not assert independent reproduction. ### Qwen3-VL-32B-Instruct Accuracy (English): 77.2. Result ID: 6a05681147f34ce924ebaac9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Supervised finetuning"}. This record does not assert independent reproduction. ### Qwen3-VL-32B-Instruct Accuracy (English): 76.9. Result ID: 6a05681147f34ce924ebaac8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Constrained decoding"}. This record does not assert independent reproduction. ### Qwen3-VL-8B-Instruct Accuracy (English): 74.1. Result ID: 6a05681147f34ce924ebaabe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Zero shot inference"}. This record does not assert independent reproduction. ### Qwen2.5-VL-32B-Instruct Accuracy (English): 74.1. Result ID: 6a05681147f34ce924ebaac6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Supervised finetuning"}. This record does not assert independent reproduction. ### Qwen2.5-VL-32B-Instruct Accuracy (English): 73.8. Result ID: 6a05681147f34ce924ebaac4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Zero shot inference"}. This record does not assert independent reproduction. ### Gemma-3-27b-it Accuracy (English): 73.7. Result ID: 6a05681147f34ce924ebaaca. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Zero shot inference"}. This record does not assert independent reproduction. ### Gemma-3-12b-it Accuracy (English): 71.7. Result ID: 6a05681147f34ce924ebaac1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Zero shot inference"}. This record does not assert independent reproduction. ### Qwen2.5-VL-32B-Instruct Accuracy (English): 70.5. Result ID: 6a05681147f34ce924ebaac5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Constrained decoding"}. This record does not assert independent reproduction. ### Qwen3-VL-8B-Instruct Accuracy (English): 69.8. Result ID: 6a05681147f34ce924ebaac0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Supervised finetuning"}. This record does not assert independent reproduction. ### Qwen3-VL-4B-Instruct Accuracy (English): 69.6. Result ID: 6a05681147f34ce924ebaab5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Zero shot inference"}. This record does not assert independent reproduction. ### Gemma-3-27b-it Accuracy (English): 69.2. Result ID: 6a05681147f34ce924ebaacb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Constrained decoding"}. This record does not assert independent reproduction. ### Qwen3-VL-8B-Instruct Accuracy (English): 69. Result ID: 6a05681147f34ce924ebaabf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Constrained decoding"}. This record does not assert independent reproduction. ### Qwen3-VL-4B-Instruct Accuracy (English): 67.7. Result ID: 6a05681147f34ce924ebaab7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Supervised finetuning"}. This record does not assert independent reproduction. ### Gemma-3-27b-it Accuracy (English): 67.7. Result ID: 6a05681147f34ce924ebaacc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Supervised finetuning"}. This record does not assert independent reproduction. ### Qwen3-VL-4B-Instruct Accuracy (English): 67.2. Result ID: 6a05681147f34ce924ebaab6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Constrained decoding"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-Instruct Accuracy (English): 63.1. Result ID: 6a05681147f34ce924ebaab3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Constrained decoding"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-Instruct Accuracy (English): 62.6. Result ID: 6a05681147f34ce924ebaab4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Supervised finetuning"}. This record does not assert independent reproduction. ### Gemma-3-12b-it Accuracy (English): 62.3. Result ID: 6a05681147f34ce924ebaac2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Constrained decoding"}. This record does not assert independent reproduction. ### Gemma-3-12b-it Accuracy (English): 62.3. Result ID: 6a05681147f34ce924ebaac3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Supervised finetuning"}. This record does not assert independent reproduction. ### Gemma-3-4b-it Accuracy (English): 61.5. Result ID: 6a05681147f34ce924ebaab8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Zero shot inference"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B-Instruct Accuracy (English): 61.2. Result ID: 6a05681147f34ce924ebaabc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Constrained decoding"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B-Instruct Accuracy (English): 61.1. Result ID: 6a05681147f34ce924ebaabd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Supervised finetuning"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-Instruct Accuracy (English): 57.5. Result ID: 6a05681147f34ce924ebaab2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Zero shot inference"}. This record does not assert independent reproduction. ### Gemma-3-4b-it Accuracy (English): 56.3. Result ID: 6a05681147f34ce924ebaab9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Constrained decoding"}. This record does not assert independent reproduction. ### Gemma-3-4b-it Accuracy (English): 56.3. Result ID: 6a05681147f34ce924ebaaba. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Supervised finetuning"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B-Instruct Accuracy (English): 55.6. Result ID: 6a05681147f34ce924ebaabb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Protocol":"Zero shot inference"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.